I am looking at it. It does not do what I want to do, but I should look at it more closely. In order to keep something like that stable, you cannot let the sentences that you input gain power over other sentences that are input unless the specifics are clearly detailed and constrained. So if you want a statement to supplement or modify a statement that was previously used you have to do so through a token reference like a name or a word or a phrase. There is not way that the program could then learn to 'think' about the sentences that are input in a creative way. Through the use of generalizations (qualified or conditional generalizations) a sentence may refer to a broader range of ideas than a narrow specification attaches some sentence to a subject. The mechanism that I have in mind would enable a sentence to have, when specified, an effect over how the program interprets other statements. The problem with this is that various statements can have unplanned effects over how the program interprets other statements and the whole purpose of using a highly constrained grammar would be defeated. At some point the unambiguous and clearly specified statements that were input would be subject to a variety of forces that could interact in unexpected ways. I think I may be able to do this though because I think I have a way to control this problem at least to some extent.
However, the mere accumulation of 'stories' concerning some subject matter can make the reference to the subject too complicated to be useful. For instance, if Bob spent the day at work then if his boss asked if he had been drinking the answer is no. However, if he took some clients out to lunch and they had a drink then the response to the boss becomes more subtle. Yes he had been drinking but he only had one drink and it was well within the nature of a customary business lunch. On the other hand if he spent the day at the bar and did no work that would be different. If a scenario dramatically changes how a specific question might be answered, it would require further insight to know how to reframe the answer so that it did not exaggerate a detail of the scenario. We would like to be able to train a computer to deal with different kinds of situations by using generalizations. But since the breadth of application of the generalizations used to train the program would have to be determined on a case by case basis this complicates the problem. If you are dealing with someone who is inquiring whether an employee has been drinking you can only figure out how to respond to questions like that on the basis of previous experiences which can be used to examine how well the outcome matched your goals. So since I want my program to be able to learn how to think from the sentences I would input, this can lead to a great many complications. However, the simple accumulation of 'stories' about some subject can in themselves lead to a great many complications even if the sentences and stories are not given greater effective power over how the program interprets information. Jim Bromer On Wed, Sep 26, 2012 at 2:49 PM, Juhani Sademaa <[email protected]> wrote: > Hi Jim > > Do you mean something like the ACE-project > Attempto Controlled English <http://attempto.ifi.uzh.ch/site/> > that uses "highly constrained forms of English sentences"? Try the editor > (in the tools). May be you could use ACE as a starting point. I do have > similar plans. > > - Juhani Sademaa - > > On Wed, Sep 26, 2012 at 2:51 AM, Jim Bromer <[email protected]> wrote: > >> I think I could write something that is a little like an active >> programming language that would be able to integrate highly >> constrained forms of English sentences into a model of the references >> of the sentences so long as the 'effects' (or local universe) of >> those references are specified by various sentences (either implicitly >> or explicitly). The constrained forms of language would hold the >> ambiguity of the parts of the sentences to a workable level. >> Jim Bromer >> >> >> >> ------------------------------------------- >> AGI >> Archives: https://www.listbox.com/member/archive/303/=now >> RSS Feed: >> https://www.listbox.com/member/archive/rss/303/13909037-423ee95b >> Modify Your Subscription: https://www.listbox.com/member/?&; >> >> Powered by Listbox: http://www.listbox.com >> > > *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/10561250-164650b2> | > Modify<https://www.listbox.com/member/?&;>Your Subscription > <http://www.listbox.com/> > ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968 Powered by Listbox: http://www.listbox.com
